Quantitation of large, middle and small hepatitis B surface proteins in HBeAg‐positive patients treated with peginterferon alfa‐2a

Hepatitis B virus (HBV) contains three viral surface proteins, large, middle and small hepatitis B surface protein (LHBs, MHBs, SHBs). Proportions of LHBs and MHBs are lower in patients with inactive vs active chronic infection. Interferon alfa may convert hepatitis B e antigen (HBeAg)‐positive chronic hepatitis B (CHB) to an inactive carrier state, but prediction of sustained response is unsatisfactory. The aim of this study was to test the hypothesis that quantification of MHBs and LHBs may allow for a better prognosis of therapeutic response than total hepatitis B surface antigen (HBsAg) concentration.


| INTRODUC TI ON
Hepatitis B virus (HBV) surface antigen (HBsAg) is essential for diagnosis of HBV infection. Monitoring of HBsAg levels is recommended in chronic hepatitis B (CHB) treatment guidelines, with HBsAg loss considered the ideal treatment outcome, commensurate to a functional cure. [1][2][3] HBsAg consists of three co-carboxyterminal proteins: large (L), middle (M) and small (S) HBs proteins ( Figure S1A). The 226 amino acid (aa)-long S domain comprises the SHBs protein and forms the carboxyterminus in LHBs and MHBs proteins. The preS2 is present in LHBs and MHBs, while the preS1 domain is present only in LHBs. 4,5 HBs proteins form the virion envelope, but also assemble into noninfectious subviral particles (SVPs) ( Figure S1B) that are secreted in approximately 3000fold excess relative to virions. 6 Most SVPs are pleomorphic 20 nm spheres with a smaller proportion of LHBs than virions.
SHBs is the major component of the virion envelope and SVPs. 5 The exterior hydrophilic loop of the multiple membrane-spanning S domain forms the major antigenic determinants of HBsAg. MHBs is nonessential for replication, 7 but is conserved in all known orthohepadnaviruses, the genus of HBV. 5 MHBs has been detected in hepatitis B e antigen (HBeAg)-positive chronic HBV carriers but was not as readily detectable in HBeAg-negative carriers, 8 Standard quantitative HBsAg assays use antibodies against the S domain and do not differentiate the three HBs proteins. 12,13 The relative abundance of these proteins was originally studied in purified SVPs and virions using gel electrophoresis. 4 Later, enzyme immune assays were used to quantify LHBs levels in a limited number of sera from HBV carriers, 8 but the true abundance of LHBs and MHBs in patients with CHB remained unclear. Recently, MHBs and LHBs were described as better predictors for a stable inactive HBV carrier state in HBeAgnegative infection than total HBsAg levels. 10 It is currently unknown whether LHBs and MHBs are also useful to more reliably predict a stable inactive HBV carrier state as sustained response to PegIFN.
The objectives of this retrospective study were to determine the levels of all three HBs proteins in a large group of HBeAg-positive patients before and during PegIFN therapy, and to assess the performance of the LHBs-, MHBs-and SHBs-specific assays in predicting HBeAg seroconversion as response to treatment with PegIFN.

| Patients
Stored serum samples from patients enrolled in two large, international, randomized phase III/IV studies (WV16240 [NCT00048945] and WV19432 [NCT00435825]) were retrospectively analysed. The design and primary results of the two studies have been published previously. 14,15 Briefly, patients had confirmed CHB, were HBeAgpositive and treatment naïve (no antiviral therapy for CHB in the previous 6 months) at screening and had elevated serum alanine aminotransferase (ALT) (>1-10 × upper limit of normal) and HBV DNA levels (>500 000 copies/mL 14 or >100 000 IU/mL). 15 Patients with decompensated liver disease, serious co-existing medical conditions or co-infection with hepatitis C or D virus or HIV were excluded.
Participants provided written informed consent including permission for future analyses of samples.
Only HBeAg-positive patients who had been randomized to PegIFN 180 μg/wk with or without lamivudine (LAM) for 48 weeks, and had available baseline and on-treatment (Weeks 12 and 24) samples were considered for this retrospective analysis. All patients had completed 24 weeks post-treatment follow-up. This study is registered at ClinicalTrials.gov: NCT01705704.

| Laboratory assessments
Clinical and laboratory data were obtained from the study spon-

Key Points
• Hepatitis B surface antigen (HBsAg) is a marker of disease and treatment response in chronic hepatitis B (CHB) and is composed of three different proteins.
• We have examined the variation in levels of different HBsAg components in patients with HBeAg-positive CHB treated with peginterferon alfa-2a, according to treatment response.
• Our findings indicate that quantification of HBs proteins may not provide additional value to predict response to peginterferon alfa-2a therapy in HBeAg-positive CHB.
• However, genotype-specific differences in HBs protein level warrant further investigation. qualitative assay using reference standards from the Paul Ehrlich Institute, Langen, Germany. HBsAg levels were measured using the Architect HBsAg (Abbott; LLOQ 0.05 IU/mL) assay in Study WV16240 and the Elecsys ® HBsAg II quant (Roche Diagnostics; LLOQ 0.05 IU/ mL) assay in Study WV19432. Serum ALT levels were measured in local laboratories and standardized using local laboratory reference ranges.
LHBs, MHBs and SHBs proteins were quantified using custom quantitative sandwich enzyme-linked immunosorbent assays (ELISAs) in microplates coated with either preS1 or preS2, or S-specific monoclonal antibodies (mabs). LHBs-specific mab MA18/7 reacts with a short peptide sequence (DPXF) in the preS1 domain. 16 MHBs-specific mab Q19/10 reacts with the N-terminal glycopeptide structure of the preS2 domain, which is not present in LHBs 17 (modified from Deepen et al; 8 ; for details see Supplementary Data). SHBs was calculated indirectly from total, LHBs and MHBs levels.

| Ethics
All studies were conducted in accordance with the Declaration of Helsinki and principles of Good Clinical Practice. The study protocols were approved by the appropriate ethics committee/institutional review board at each participating centre, as detailed in the supplementary material.

| Patient characteristics
For this analysis, the cohort of 127 patients with stored serum samples available represented a small (14%) responder-enriched subpopulation of 901 patients who received PegIFN with or without LAM for 48 weeks in the two clinical trials. 14,15 In this cohort, 74 had received PegIFN and 53 received PegIFN plus LAM, with no notable differences between the two subgroups (Table 1). Most patients were young (mean age: 30 years), male, of East Asian ethnicity, and had HBV genotype B (35%) or C (61%) infection. Compared with genotype C, genotype B infection was associated with significantly higher HBV DNA, ALT and HBsAg levels (
Levels of all three HBs proteins correlated strongly with quantitative HBsAg (r s = 0.94-0.97, P < .0001) and moderately with HBeAg and HBV DNA (r s = 0.53-0.74, P < .0001), but not with age or ALT levels (r s < 0.25; Figure 2A). Overall, similar correlations were observed for total HBsAg and HBs protein proportions ( Figure 2B).

TA B L E 1 Summary of patient characteristics
and MHBs (4% vs 3%, P = .0428) when compared with genotype C ( Figure 3B). The higher proportion of LHBs was confirmed by Western blot analysis conducted in a randomly selected subset of patients (n = 41 genotype B, n = 21 genotype C), using a mab against a linear epitope of the S domain. The Western blot analyses revealed much higher MHBs levels (21% genotype B, 17% genotype C) than seen in ELISA leading to a slightly lower proportion of SHBs (72% and 80%; Figure 3C). ELISA results for the subset whose samples were subjected to Western blot analysis were comparable with the overall population. The difference in HBsAg composition between genotype B and C subgroups persisted during PegIFN ± LAM treatment ( Figure S2B).

| Hepatitis B surface protein levels during peginterferon alfa-2a therapy
Individual HBs protein levels decreased steadily during the initial 24 weeks of treatment (~1 log decline in mean levels by Week 24) ( Figure 4A). A moderate change was observed in the composition of HBs protein proportions; by Week 24, MHBs and LHBs had slightly decreased (from 4% to 2%, and from 8% to 6% respectively), while the proportion of SHBs had slightly increased (from 88% to 92%; Figure 4B).
Concomitant administration of LAM did not significantly affect individual HBs protein levels during treatment ( Figure 4C; P > .05 for between-group comparisons at each time point, unpaired t test), and this was also the case for ALT, HBsAg and HBeAg levels ( Figure S3A).
In responders with HBeAg seroconversion at 24 weeks post-treatment, mean HBs protein levels were significantly lower than in nonresponders at all time points ( Figure 4D; P < .05 for every comparison). Likewise, the mean HBeAg, HBsAg and HBV DNA levels, but not ALT levels, were lower in responders than nonresponders at all time points ( Figure S3B).

| Baseline and on-treatment predictors of response
Given that both treatment subgroups had comparable response rates and on-treatment biomarker kinetics (apart from HBV DNA), biomarker data were pooled for further analysis. Pooling of treatment subgroup data was not deemed appropriate for HBV DNA; thus, further analyses were restricted to the PegIFN monotherapy subgroup.
In addition, individual biomarker cut-offs were further explored to assess their utility for use as stopping rules early during treatment.
Receiver operating characteristic curve analyses of biomarker data ( Figures S4 and S5)  Individual biomarker cut-offs were explored for absolute biomarker levels (higher ROC AUCs than change from baseline), that were associated with high negative predictive values (90%) at Week 12 of treatment. Cut-offs were identified for HBV DNA, HBeAg and HBsAg, as well as for the S/M/LHBs proteins, but not for ALT levels.
The LHBs (960 ng/mL) and quantitative HBsAg (27 000 IU/mL) cutoffs identified the largest proportions of nonresponders (32% and 31% respectively), followed by the HBV DNA (8.9 log 10 IU/mL, 26%), SHBs (27 000 ng/mL, 18%), MHBs (1500 ng/mL, 16%) and HBeAg (1700 IU/mL, 12%) cut-offs. It is noteworthy that the ELISA and Western blot results for LHBs -in contrast to those for MHBs -agreed very well. This suggests that the ratio of internal and external preS1 domains in virions or HBsAg filaments is not highly variable, and that external preS1   Figure 3C). This study has certain limitations. We evaluated HBs levels retrospectively using stored serum samples from a subset of patients enrolled in a randomized controlled trial. Thus, it is possible that the results have been affected by sample deterioration or selection bias. Our study cohort was found to be responder-enriched,  10 where genotype B and C were not the most common. The population was also predominantly of East Asian origin; future analyses should strive to include a more ethnically diverse population. Finally, the underestimation of MHBs in ELISA is an additional limitation of this analysis.

| D ISCUSS I ON
In conclusion, the results of this retrospective analysis show that SHBs levels are strongly correlated with quantitative HBsAg levels; and that HBs protein levels differ by HBV genotype. SHBs/ MHBs/LHBs protein levels were found to be significantly lower in responders than in nonresponders during PegIFN treatment, and to predict PegIFN response in a manner comparable to quantitative HBsAg levels. Although HBs proteins were not superior to total HBsAg for predicting response to PegIFN, LHBs and MHBs represent novel biomarkers that warrant further investigation.